I think you're partially right about the reasons (the other half of the equation simply being the momentum of Python 2 efforts), but I think your last clause ("

I think you're partially right about the reasons (the other half of the equation simply being the momentum of Python 2 efforts), but I think your last clause ("it's just useless") is incorrect. I can tell you for 100% certain that, if a project of mine could run on a version of "Python" that was even twice as fast, I would switch immediately, because 6 application servers could become 3. If the gains are even greater, that's tremendous. It's not just scientific computing that benefits, and in fact, a lot of the scientific / numerical analysis done in Python is really done in C / C ++ anyway (think Numpy, et al which delegate to compiled code for much of their expensive operations).